The only method change was the shared #moveChangesTo: in TFileInOutDescription, as described in the report on Mantis. There was a bit of manual work involved in fixing up the Traits in the image though (just as part of the investigation into the problem). For instance, some methods which were not local to particular traits and classes had their own compiled method instances, even without super sends.
These problems were, as it turns out, probably irrelvant. The main problem was just the fact that the source for the shared compiled methods was written to the new changes multiple times while updating the source pointer in the compiled method.
-----Original Message----- From: squeak-dev-bounces@lists.squeakfoundation.org [mailto:squeak-dev-bounces@lists.squeakfoundation.org]On Behalf Of Dan Ingalls Sent: 30 May 2006 10:44 PM To: The general-purpose Squeak developers list Subject: RE: condenseChanges broken in 3.9
Change set might be awkward since the changes were condensed!
Well, it sounds that way, but condenseChanges should never affect the most recent version, which is all that is needed for fileOut of the changeSet. Did you really mean there's a problem?
- D
-----Original Message----- From: squeak-dev-bounces@lists.squeakfoundation.org [mailto:squeak-dev-bounces@lists.squeakfoundation.org]On Behalf Of stephane ducasse Sent: 30 May 2006 10:45 AM To: The general-purpose Squeak developers list Subject: Re: condenseChanges broken in 3.9
I'm going through your discussion and it would be nice if we could get one or two tests at least from the queries you did so that we know that this is really fixed.
Stef
On 30 mai 06, at 11:23, Gary Chambers wrote:
Hi all, managed to fix condenseChanges finally.